An extensive body of research exists to support the involvement of a multistep process in the conversion of normal cells to the tumorigenic phenotype (see, e.g., Land et al., 1983). Molecular models supporting this hypothesis were first provided by studies on two DNA tumor viruses, adenovirus and polyomavirus. In the case of adenovirus, it was found that transformation of primary cells required the expression of both the early region 1A (E1A) and 1B (E1B) genes (Houweling et al., 1980). It was later found that the E1A gene products could cooperate with middle T antigen or with activated H-ras gene to transform primary cells (Ruley, 1985). In addition, during the last decade, a number of human malignancies have been discovered to be correlated with the presence and expression of "oncogenes" in the human genome. More than twenty different oncogenes have now been implicated in tumorigenesis, and are thought to play a direct role in human cancer (Weinberg, 1985). Many of these oncogenes apparently evolve through mutagenesis of a normal cellular counterpart, termed a "proto-oncogene", which leads to either an altered expression or activity of the expression product. There is considerable data linking proto-oncogenes to cell growth, including their expression in response to certain proliferation signals (see, e.g., Campisi et al., 1983) and expression during embryonic development (Muller et al., 1982). Moreover, a number of the proto-oncogenes are related to either a growth factor or a growth factor receptor. These observations suggested the involvement of multiple functions in the transformation process, and that various oncogenes may express similar functions on a cellular level.
The adenovirus E1A gene codes for several related proteins to which a number of interesting properties have been attributed. In addition to its ability to complement a second oncogene in transformation, a closely related function allows E1A to immortalize primary cells (Ruley, 1985). For example, introduction of E1A gene products into primary cells has been shown to provide these cells with an unlimited proliferative capacity when cultured in the presence of serum. Another interesting action of E1A function is so-called "trans-activation", wherein E1A gene products stimulate transcription from a variety of viral and cellular promoters, including the adenovirus early and major late promoter as well as other promoters. However, trans-activation is not universal for all promoters. In some instances, E1A causes a decrease in transcription from cellular promoters that are linked to enhancer elements (Haley et al., 1984). It has been shown that exogenously added E1A gene can reduce the metastatic potential of ras-transformed rat embryo fibroblast cells by activating the cellular NM23 gene that is associated with a lower metastatic potential (Pozzatti et al., 1988; Wallich et al., 1985).
In the case of Adenovirus 5, the E1A gene products are referred to as the 13S and 12S products, in reference to the sedimentation value of two mRNAs produced by the gene. These two mRNAs arise through differential splicing of a common precursor, and code for related proteins of 289 and 243 amino acids, respectively. The proteins differ internally by 46 amino acids that are unique to the 13S protein. A number of E1A protein species can be resolved by PAGE analysis, and presumably arise as a result of extensive post-translational modification of the primary translation products (Harlow et al., 1985).
Another viral oncoprotein, the SV 40 large T antigen (LT) shares structural and functional homology to E1A and c-myc (Figge et al., 1988). LT, E1A and c-myc have transforming domains which share amino acid sequence homology and similar secondary structure (Figge et al., 1988). All three proteins complex with the tumor suppressor, retinoblastoma gene product (Rb) (Whyte et al., 1988, DeCaprio et al., 1988, Rustgi et al., 1991), and the Rb binding domains of LT and E1A coincide with their transforming domains. Based on this similarity, it has been thought that LT and E1A transform cells by binding cellular Rb and abrogating its tumor suppressor function. LT, E1A and c-myc are also grouped as immortalization oncogenes as determined by the oncogene cooperation assay using rat embryo fibroblasts (Weinberg, 1985).
In spite of the similarity between the Rb binding domains of LT and E1A, the two proteins differ substantially in other regards. In fact, there is apparently only a short equivalent stretch of acidic amino acids (Figge et al., 1988). This stretch lies between amino acids 106-114 in LT and amino acids 121-139 in E1A. The large T antigen is encoded by the simian virus 40, a member of the polyoma virus family. In contrast, E1A is encoded by adenovirus 5 virus, which is a member of the adenovirus family. LT is 708 amino acids long, while E1A is substantially shorter at 289 or 243 amino acids (for 13S and 12S respectively). LT has been observed to bind directly to certain DNA sequences; however, direct binding of E1A to DNA has not been observed and E1A may instead interact indirectly via a co-activator such as p300 (as discussed in Chen & Hung, 1997). LT binds with the tumor suppressors Rb and also with p53. E1A complexes with Rb but apparently not with p.sup.53. E1A has been shown to induce apoptosis in cells, but this has not been demonstrated for LT. Further, LT is an apparent anomaly in the scheme of oncogenic classification. Oncogenes are typically classified as being cytoplasmic or nuclear oncogenes. However, LT, through the actions of a single protein, is able to introduce "nuclear" characteristics such as immortalization as well as "cytoplasmic" characteristics such as anchorage independence in cells (Weinberg, 1985). LT antigen can be found in both the nucleus and at the plasma membrane, and mutations that inhibit the transport of LT into the nucleus appear to reduce its immortalizing ability while leaving intact its effect on anchorage independence and its ability to transform already immortalized cells. Consequently, this oncogene is considered to be a member of both the nuclear and cytoplasmic oncogenic classes, since it its gene product apparently affects these two distinct cellular sites (Weinberg, 1985); which again is unlike E1A.
Despite advances in identifying certain components which contribute to the development of malignancies, it is clear that the art still lacks effective means of suppressing carcinogenesis. Recently, however, M. C. Hung and collaborators have made great advances in the suppression of oncogenic transformation. Some of these advances are described in U.S. Pat. Nos. 5,651,964, 5,641,484, and 5,643,567, the entire text of each being specifically incorporated by reference herein and briefly described below.
Suppression of Oncogenesis
Work by Hung and collaborators has established that the E1A gene can in fact suppress transformation, tumorigenicity and metastasis in a variety of cancers (see, e.g., Yu et al. 1991, 1992 and 1993; and the reviews by Hung et al., 1995, Yu and Hung, 1995, and Mymryk, 1996).
Without wishing to be bound by theory, it appears that there may be more than one pathway or means by which E1A can act to suppress oncogenic transformation. In particular, while Hung and collaborators initially established that E1A can suppress tumor formation in vitro and in vivo in cancers that appear to be associated with an over-expression of an oncogene variously referred to as c-erbB-2, HER-2 or neu (hereinafter the neu oncogene); it appears that E1A can also suppress the oncogenic phenotype in various other cancer cells that do not appear to be associated with an overexpression of neu. Indeed, Frisch et al. have reported that the tumor-suppressing effects of the E1A gene can also be used to convert three unrelated types of human cancer cells (which do not appear to be over-expressing neu) into a non-transformed state (see, e.g., Frisch, 1991, 1994 and 1995, and Mymryk, 1996). However, even these cells appear to express neu at some relatively lower levels. Whether the suppression of transformation in cells not over-expressing neu is nevertheless facilitated by a reduction in neu levels and/or is facilitated by other means, the outcome in either case is suppression of oncogenesis. In sum, therefore, it appears that E1A can effectively function as a tumor suppressor gene for a variety of different human cancer cells including both cancer cells that are overexpressing neu, and those that are not. E1A protein has also been reported to induce a cytotoxic response that resembles programmed cell death (i.e. apoptosis) (Rao et al., 1992), which may also contribute to the tumor-suppressing properties of E1A.
These results not only establish E1A as a tumor suppressor gene, but also suggest that E1A is a potential therapeutic reagent for the treatment of a variety of human cancers. Indeed, success with the use of E1A as a tumor suppressor gene in animal models of human cancer has merited the initiation of Phase I human clinical trials for multiple indications which are currently being sponsored by Targeted Genetics Corporation at the Virginia Mason Medical Center in Seattle, at the M. D. Anderson Cancer Center in Houston, and at Wayne State University in Detroit, at the Rush Presbyterian--St. Luke's Medical Center in Chicago. In addition, Targeted Genetics' European partner, Groupe Fournier, has now received approval by the Ministry of Health to begin corresponding clinical trials in France.
As noted above, one of the ways in which E1A may mediate suppression of the oncogenic phenotype is through an effect on the putative oncogene c-erbB-2/HER-2/neu, overexpression of which is associated with a variety of human cancers, including human breast and ovarian cancers among others. The c-erbB-2/HER-2/neu oncogene has been found to be similar to, but distinct from, the c-erbB gene, which is a member of the tyrosine-specific protein kinase family to which many proto-oncogenes belong. The c-erbB gene encodes the epidermal growth factor receptor (EGFr) and is highly homologous to the transforming gene of the avian erythroblastosis virus (Downward et al., 1984).
The neu oncogene, which encodes a p185 tumor antigen, was first identified in transfection studies in which NIH 3T3 cells were transfected with DNA from chemically induced rat neuroglioblastomas (Shih et al., 1981). The p185 protein has an extracellular, transmembrane, and intracellular domain, and therefore has a structure consistent with that of a growth factor receptor (Schechter et al., 1984). The human neu gene was first isolated due to its homology with v-erbB and EGF-r probes (Senba et al., 1985). Molecular cloning of the transforming neu oncogene and its normal cellular counterpart, the neu proto-oncogene, indicated that activation of the neu oncogene was due to a single point mutation resulting from one amino acid change in the transmembrane domain of the neu encoded p185 protein (Bargmann et al., 1986; Hung et al., 1989). The neu oncogene is of particular importance to medical science because its presence has been correlated with the incidence of cancers of the human breast and female genital tract among others. Moreover, amplification/overexpression of this gene has been directly correlated with relapse and survival in human breast cancer (Slamon et al., 1987). Therefore, it is an extremely important goal of medical science to evolve information regarding the neu oncogene, particularly information that could be applied to reversing or suppressing the oncogenic progression that seems to be elicited by the presence or activation of this gene. Unfortunately, little has been previously known about the manner in which one may proceed to suppress the oncogenic phenotype associated with the presence of oncogenes such as the neu oncogene.
The neu proto-oncogene is often notably amplified in patients with metastatic breast cancer. Hung et al. have shown that neu transcription can be repressed by E1A products in an established rat embryo fibroblast cell line, Rat-1. Furthermore, Hung et al. have found that in SK-BR-3 human breast cancer cells expression of the p185 protein, the human neu gene product, was reduced by introduction of E1A gene. The depression effect observed in the co-transfection experiment with the Stu1-Xho1 fragment has demonstrated that this reduction of p185 proteins is likely due to the similar transcriptional repression mechanisms.
As noted above, Hung and collaborators, who have been studying various cancers that appear to be associated with over-expression of the neu oncogene, have successfully demonstrated that E1A gene products are able to suppress not only the tumorigenic and transformation events but are also able to suppress metastatic events associated with such cancers. See, e.g., Yu et al., 1992; Yu et al., 1991; Yu et al., 1993. As described by Yu et al., 1993, SKOV3.ip1 is a derivative cell line isolated from the ascites that developed in mice given injections of human ovarian carcinoma SKOV-3 cells. Compared with parental SKOV-3 cells, the SKOV3.ip1 cell line expresses higher levels of c-erbB-2/neu-encoded p185 protein and correspondingly exhibits more malignant phenotypes determined by in vitro and in vivo assays. This association between enhanced c-erbB-2/neu expression and more severe malignancy is very consistent with previous studies in which c-erbB-2/neu overexpression was shown to correlate with poor prognosis in ovarian cancer patients (Slamon et al., 1989). These studies provided actual evidence to support those clinical studies indicating that c-erbB-2/neu overexpression can be used as a prognostic factor for ovarian cancer patients and that c-erbB-2/neu overexpression may play an important role in the pathogenesis of certain human malignancies such as ovarian cancer. The identification and molecular cloning of the ligands for the c-erbB-2/neu-encoded p185, which can increase the tyrosine phosphorylation of p185, will enhance our understanding of the molecular mechanisms and the biological effects of c-erbB-2/neu overexpression in human cancer and cancer metastasis (Peles et al., 1992; Holmes et al., 1992; Lupu et al., 1990; Yarden & Peles, 1991; Huang & Huang, 1992; Dobashi et al., 1991).
The adenovirus E1A gene was originally defined as a transforming oncogene that can substitute for the myc oncogene and simian virus 40 large tumor antigen gene in the ras co-transformation assay of primary embryo fibroblasts (Land et al., 1983; Ruley, 1983; Weinberg, 1985).
As noted above, Hung et al. discovered that E1A gene products can act as transformation and metastasis suppressors in transformed mouse 3T3 cells. It was further demonstrated that the E1A gene products effectively repressed c-erbB-2/neu gene expression in SKOV3.ip1 ovarian carcinoma cells, suppressed transformation phenotypes in vitro, and reduced tumorigenicity and mortality rate in vivo. Hence it was demonstrated that the adenovirus E1A gene can function as a tumor suppressor gene for human cancer cells as well as inhibit transformation induced by a mutation-activated neu oncogene in rodent cells. Without wishing to be bound by theory, it appears that the reduced p185 expression in the ip1.E1A cell lines may be due to transcriptional repression of the overexpressed c-erbB-2/neu gene, which may be one of the means by which E1A can suppress the tumorigenic potential of SKOV3.ip1 ovarian cancer cells. Interestingly, it has been shown that adenovirus E1A can also render hamster cell lines more susceptible to lysis by natural killer cells and macrophages (Cook & Lewis, 1984; Sawada et al., 1985); and it increased sensitivity to cytotoxicity by tumor necrosis factor in transfected NIH 3T3 cells (Cook et al., 1989). Therefore, it is conceivable that the tumor-suppressing function of E1A may be partly due to an increased susceptibility to cytolytic lymphoid cells and molecules.
It has been proposed that there are cellular "E1A-like" factors that may mimic the function of E1A in certain cell types (Nelson et al., 1990). Many common features between E1A and c-myc suggest that the c-myc gene product may be one of the cellular homologues of the E1A protein. These common features include the following: E1A and c-myc share a similar structural motif (Figge & Smith, 1988; Figge et al., 1988); both E1A and c-myc can transform primary embryo fibroblasts in cooperation the ras oncogene (Land et al., 1983; Ruley, 1983); both can bind specifically to the human Rb gene product, the Rb protein (Whyte et al., 1988; Rustgi et al., 1991); both can induce apoptosis in certain cell types (Rao et al., 1992; Frisch, 1991; Nelson et al., 1990; Figge & Smith, 1988; Figge et al., 1988; Whyte et al., 1988; Rustgi et al., 1991; Evan et al., 1992); and both have been shown to block transformation of certain transformed cell lines (Frisch, 1991; Nelson et al., 1990; Figge & Smith, 1988; Figge et al., 1988; Whyte et al., 1988; Rustgi et al., 1991; Evan et al., 1992; Suen & Hung, 1991). In addition, Hung et al. have found that, similar to the E1A proteins, the c-myc gene product can repress c-erbB-2/neu gene expression at the transcription level, resulting in reversal of the neu-induced transformed morphology in NIH 3T3 cells (Wang et al., 1991).
It appears that E1A can inactivate the Rb tumor suppressor gene by complexing with the Rb gene product, Rb protein, and by inducing Rb protein phosphorylation (Whyte et al., 1988; Rustgi et al., 1991; Evan et al., 1992; Suen & Hung, 1991; Wang et al., 1991). Therefore, Hung et al. have recently examined whether Rb might also regulate c-erbB-2/neu expression. Similar to E1A, Rb can also repress c-erbB-2/neu gene expression at the transcriptional level (Yu et al., 1992). The cis-acting elements responding to E1A and Rb are different but only a few base pairs away from each other. It may be that E1A and Rb might interact with each other to regulate c-erbB-2/neu transcription.
Without wishing to be bound by theory, one of the interesting issues regarding the correlation between c-erbB-2/neu overexpression and poor clinical outcome in human breast and ovarian cancers is whether c-erbB-2/neu overexpression is the result of an aggressive tumor or has a causative role for aggressive tumors. The data presented by Hung et al. (U.S. Pat. Nos. 5,651,964, 5,641,484, and 5,643,567) supported a role for c-erbB-2/neu overexpression in the pathogenesis of certain aggressive tumors. First, comparison of the SKOV-3 cell line and the derivative SKOV3.ip1 cell line revealed a correlation between increased c-erbB-2/neu expression level and enhanced malignant phenotype measured by in vitro and in vivo assays. Second, when c-erbB-2/neu expression in the E1A-expressing ip1.E1A cells was dramatically repressed, the malignant potential of these cells was diminished. Taken together, these observations suggest a close relationship between c-erbB-2/neu over-expression and the more malignant tumor pattern. Since tumorigenesis is likely to be a multi-step process, as noted above, it is also possible that the neu oncogene contributes to the development or progression of a tumorigenic phenotype in certain cancers even if it does not initiate the process. In that regard, neu can also serve as an indicator of the state of tumorigenesis. For example, c-erbB-2/neu-overexpressing ovarian tumors tend to be more malignant, and therefore more aggressive therapy might be beneficial to those ovarian cancer patients whose tumors overexpress c-erbB-2/neu-encoded p185. As noted above, one of the ways in which E1A might suppress tumorigenesis is via c-erbB-2/neu, which could involve indirect control at the transcriptional level. It has been proposed that E1A may form a complex with cellular transcription factor(s) and thereby modulate the specific binding of the transcription factor(s) to enhancer elements that are important for transcription (Mitchell et al., 1989). Identification of the defined DNA sequences responsible for the E1A-mediated inhibition of neu transcription would therefore allow the identification of the transcription factor(s) that may be involved in this process. Recent work by Chen and Hung provides evidence that p300 may act as such a co-activator in the transcriptional regulation of neu (Chen and Hung, 1997).
Regardless of the precise mechanism of action, the work of Hung and collaborators (using neu-overexpressing cancer cells), taken in conjunction with the work by Frisch et al. (using non-neu-overexpressing cancer cells), provide evidence that E1A can effectively function as a tumor suppressor gene for a variety of different human cancer cells including cancer cells that are overexpressing neu, and those which are not. E1A has also been shown to sensitize cancer cells to chemotherapeutic agents and thus can be used as a combination therapy in the treatment of cancer cells. This tumor sensitization effect of E1A also appears to be active in both neu-over-expressing cancer cells and non-neu-over-expressing cancer cells as shown by Hung and collaborators, and by Frisch et al. (see, e.g., PCT/US97/03830 and PCT/US95/11342).
While the role of E1A as a tumor suppressor gene has thus been established, it has not been clear which regions of E1A are required for this suppression. The present invention describes portions of E1A that are apparently necessary for the tumor suppressor activity and those regions that are apparently dispensable. Removal of various portions of E1A has resulted in the generation of various "mini-E1A" genes that can be used as alternative means for providing the E1A tumor suppressor activity. Such mini-E1A genes will be useful as potential therapeutic reagents for the treatment of various human cancers.